Juvenile transfer status and the sentencing of violent offenders: a test of the liberation hypothesis

ABSTRACT

A small but growing body of research has explored how juveniles transferred to the criminal court are sentenced relative to adult defendants, but the findings from this literature have been complex and inconsistent. A noteworthy line of inquiry that to date has received only limited attention is how crime type might moderate these relationships. Theoretically, according to the liberation hypothesis, primary offense type corresponds closely with the exercise of judicial discretion, and court actors’ decision-making is most likely to be informed by extralegal offender-based attributions in the disposition of less serious cases. The goal of the present study is to extend this literature by exploring the main and interactive effects of juvenile status and crime type on adult court punishment outcomes among defendants sentenced for seven violent felony offenses. Using data from Florida circuit courts (N = 198,362), the findings show that, regarding sentencing to prison, transferred youth are consistently punished more severely than adult defendants among the three least serious crime types. However, juveniles receive shorter prison terms than adults for most violent offenses, and these disparities are greatest among murder, manslaughter, and robbery/carjacking cases.

KEYWORDS:

• Juvenile transfer
• sentencing
• crime type

Acknowledgments

The author would like to thank William D. Bales for providing the data that were used to conduct this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. Of the 5,068 murder cases, 35 offenders received jail sentences, with only 1 of these in the ages 14-17 subgroup and 3 in the ages 18-20 subgroup. Likewise, among the 3,107 sex offenders, 35 received jail sentences, including 4 defendants ages 14-17 and 7 who were ages 18-20. Similar small counts also are seen among manslaughter and ‘other’ violent offense cases, especially within the transferred juvenile subsample.

2. The prison sentence length measure has a range of 12 to 600 months. The median value is 53 months, and the modal sentence length (9.2% of the sample) is 60 months. The standard deviation of this variable is 135.1, thus indicating heavy positive skewness.

3. Supplementary analyses were conducted that made use of this continuous measure of age along with a quadratic transformation of the variable to capture nonlinearity. The findings from these models were substantively identical to those presented, and they suggest that little substantive variation in the effects of age is lost through the creation of these four categories. Additional analyses making use of more precise age categories among older offenders (i.e., ages 30-39, 40-49, 50-59, and 60+) also were conducted; however, these latter estimates were severely compromised by the very small number of defendants ages 50 and older who committed certain violent offenses (e.g., sex offenses, robbery/carjacking, and resisting arrest with violence), with even fewer receiving sentences to prison.

4. The ‘other’ violent offense category includes crimes for which the subsample sizes are too small to generate meaningful estimates for age. These include, for example, child abuse/neglect, aggravated stalking, arson, kidnapping, abuse/neglect of an elderly person or disabled adult, extortion, false imprisonment, and making a false report of a bomb threat.

5. Asian, Native American, and Pacific Islander defendants were excluded from the original dataset by the FDOC data administrator due to their very small sample size (less than 1%).

6. The variance inflation factor (VIF) values indicate that there are no issues of multicollinearity among any of the independent variables in these analyses.

7. Robust standard errors for both portions of the model are calculated using the suest command in Stata 16.1 as directed by Long and Freese (2014, 529-530) for estimating hurdle models.

8. Scholars debate whether average marginal effects (AMEs) or marginal effects at the means (MEMs) provide better estimates of the differences in the predicted values (see, e.g., Long and Freese 2014; Williams 2012). On one hand, because the calculation of MEMs involves fixing the values of the covariates at their means, they produce estimates of the marginal effects for cases that arguably could be considered typical. However, the predicted values at the means of certain variables (e.g., gender) may not be intuitively interpretable. In contrast, AMEs involve comparisons between hypothetical populations with the same observed values on the covariates except the variable of interest, thus making use of all of the available data rather than just the variables’ means. While the AMEs are presented, the MEMs also were calculated, and they revealed the same substantive pattern as that shown.

Additional information

Notes on contributors

Peter S. Lehmann

Peter S. Lehmann is an assistant professor in the Department of Criminal Justice and Criminology at Sam Houston State University. His research interests include juvenile justice and delinquency, racial and ethnic disparities in punishment, school discipline and safety, and public opinion on crime and criminal justice policy. His recently published work has appeared in Justice Quarterly, Crime & Delinquency, Punishment & Society, Youth Violence and Juvenile Justice, and other journals.